Workday CTO Peter Bailis Leaves for Anthropic to Lead Reinforcement Learning Engineering
Key Points
- Peter Bailis left his role as Workday CTO in March 2026.
- He joined Anthropic as a member of technical staff, focusing on reinforcement learning engineering.
- Anthropic’s MTS title reflects a flat technical hierarchy used at leading AI labs.
- The hire aligns with Anthropic’s aggressive push into enterprise markets and HR applications.
- Anthropic reports a revenue run rate above $30 billion and an enterprise customer base exceeding 1,000 firms.
- Workday has not announced a replacement for Bailis and continues to develop its Agent Builder tools.
- The move illustrates a growing trend of senior tech executives trading C‑suite roles for hands‑on AI research positions.
Peter Bailis, who joined Workday as chief technology officer in May 2025, departed the enterprise‑software firm last month to become a member of technical staff at AI startup Anthropic. The move trades a C‑suite title for a hands‑on role focused on reinforcement learning engineering, giving Bailis direct access to cutting‑edge model training while placing him inside a company that is now building the kind of HR software Workday sells. Anthropic’s rapid push into enterprise markets and its $30 billion revenue run rate make the hire a strategic signal for both firms.
Peter Bailis, who spent the past several months as chief technology officer of Workday, left the $8 billion‑revenue software maker in March and joined Anthropic as a member of technical staff. The shift removes a C‑suite title but places him at the front lines of reinforcement learning engineering, a core technology behind Anthropic’s Claude models.
From C‑suite to individual contributor
At Workday, Bailis oversaw a technology organization of roughly 18,000 employees and steered the company’s agentic AI strategy. Before that he led AI‑for‑data initiatives at Google Cloud, founded decision‑intelligence startup Sisu Data, and taught computer science at Stanford. Anthropic’s "member of technical staff" (MTS) designation mirrors the flat hierarchy used at OpenAI, where senior engineers and researchers share the same title. Compensation for MTS roles ranges from $300,000 to $405,000 in base salary at Anthropic, with equity pushing total pay into the seven‑figure range for senior staff.
The trade‑off is explicit: Bailis gives up institutional authority for direct involvement in model training, data pipelines, and research agendas. Anthropic, now reporting a revenue run rate above $30 billion and serving more than 1,000 enterprise customers that each spend over $1 million annually, is no longer a pure research lab. The scale of engineering challenges it faces—especially around reinforcement learning from human feedback—appeals to a technically ambitious executive.
Anthropic’s enterprise push
Anthropic’s recent moves underscore its ambition to become a platform inside large organizations. In early March the company launched a Claude‑powered marketplace that lets firms with committed API spend buy third‑party applications without the typical cloud‑provider revenue cut. Partners such as Snowflake, legal‑AI firm Harvey, and developer platform Replit were among the first to list tools.
Days later Anthropic announced a $100 million Claude Partner Network anchored by consulting giants Accenture, Deloitte, Cognizant and Infosys. The network is designed to accelerate Claude deployments across the world’s biggest enterprises, and Anthropic plans to increase the commitment in coming years while scaling partner‑facing headcount fivefold. A separate joint‑venture discussion with private‑equity firms, including Blackstone, could embed Claude across portfolio companies in exchange for roughly $200 million of Anthropic capital.
Enterprise customers now account for about 80 % of Anthropic’s revenue, and the company positions its offering as a platform that operates alongside existing workflows rather than replacing them. In practice, Claude’s ability to generate job descriptions, onboarding packs and offer letters blurs that line, bringing it into direct competition with traditional HR software.
Bailis’s background makes him a uniquely valuable addition. His tenure at Google Cloud involved building AI products that integrated with structured enterprise data, and at Workday he led the AI strategy for HR and finance applications. Anthropic has already launched Claude plugins for HR use cases, and Bailis will help translate his deep knowledge of enterprise stacks into product roadmaps that target the same budgets Workday once serviced.
The hire also highlights a broader talent trend: senior executives from established tech firms are increasingly swapping corporate titles for individual‑contributor roles at frontier AI labs. Observers note that the motivation is often the belief that the most consequential technical work now happens inside AI research organizations. For companies like Workday, losing a CTO to a direct competitor that also uses its own software creates a competitive irony. Workday’s CEO Carl Eschenbach confirmed earlier this year that Anthropic, Google and OpenAI all run Workday’s HR platform internally.
Workday has not yet named a replacement for Bailis. The firm continues to roll out its Agent Builder tools, which let customers build AI agents on top of Workday data, positioning the product as a differentiated alternative to Anthropic’s Claude‑based solutions. Whether the CTO vacancy will be filled quickly or remain open as the company retools its AI strategy remains to be seen.